REGRESSION ARTIFACTS IN TIME-SERIES AND LONGITUDINAL DATA

The problem of ''regression artifacts'' in causal inference, otherwise known as the problem of en or and reliable irrelevant variance in ''i ndependent'' variables used for matching or regression adjustment, is illustrated first in the time-series data where a treatment is trigger ed by an extreme measure. The ''offset effect'' in psychotherapy, Unde rwood's scalloped learning curve, and potential pseudo-effects in AIDS therapies are used as illustrations. The magnitude of such artifacts is computable if the autocorrelation pattern for various lags is known , and thus could be distinguished from genuine effects. For longitudin al studies in which a population of respondents is repeatedly measured , the problem of anchoring the matching or regression adjustments on a single wave of measurement (usually the first) is illustrated as affe cted by the proximally autocorrelated nature of such measures. Data fr om a famous study of the effects of job training are reinterpreted in light of this consideration. Copyright (C) 1996 Elsevier Science Ltd.